# Harberger tax on validator spots * Significantly increase (or get rid of?) `MAX_EFFECTIVE_BALANCE`, allow large stakers to consolidate * Limit active validator count to $2^{19}$ or $2^{20}$, or however many attestations per slot the beacon chain can handle for SSF * When the count is reached, activate harberger tax/auction: * Each operator decides how much their spot as a validator is worth, in ETH * Anyone can take over an operator's validator spot by paying them their listed price * Each operator pays a % of their spot's self-valuation as a recurring fee to keep their spot active (the fee gets burned, of course) * **Results:** * ✅ Net staking yield (and overall ETH issuance) quickly approaches marginal cost-of-money + operational costs for each individual validator, without needing any funky reward curve manipulation * Basically, the harberger tax captures the economic utility of each validator spot, in a race-to-the-bottom fashion * ✅ Centralized staking-as-a-service operations are disincentivized, as their 10-20% fee on their customers' yield means significantly less economic utility per validator for the actual operators, compared to someone willing to stake the same amount of their own ETH and getting 100% of the staking yield from it * ❌ Validators with small amounts of ETH are kicked out first by larger players, average ETH per validator would trend upwards * However, they do receive the agreed-upon payout when it happens – basically an upfront "severence" of x weeks/months' worth of yield, paid by the person who kicked them out * Decentralized staking pools and/or DVTs can help the access to beacon chain validation to small stakers and result in 1 spot with many ETH being shared by many distinct entities * :information_source: There's no reason to hoard spots by splitting a large stake into many validators – the total harberger fees you pay will effectively scale with the total yield received across all the validators you own. * In fact, it's better for everyone when large stakes are consolidated into single validators, because the auction is paused when there are free validator spots available. Hopefully reaching a utopia where 1 validator = 1 or more distinct entities, compared to today's paradigm where there are ~10k distinct entities for 600k+ validators * :question: Unclear interaction with external yield, e.g. Eigenlayer ## Rationale For simplicity, say the staking yield is 5% and the harberger fee is 10% of your self-valuation yearly: A small validator with 10 ETH will make 0.5 ETH/year, so he won't value his spot at more than 5 ETH, otherwise he will pay more in fees than he'll earn in yield. If he does get paid out 5 ETH then that's 10 whole years of staking yield at once. A larger validator with 100 ETH will similarly not value the spot at more than 50 ETH, though he will happily price it at 5 ETH knowing that the smaller validator will be bought off first. Maybe then deposit sizes do need to be equal across the board? Or there has to be a mechanism that would force the self-valuation to be proportional with the deposit size somehow Analogies with real world land don't really work, the validator spot of a small staker would be akin to a small plot of land, but a large staker could take over and make the land bigger somehow. Maybe we need to think in different terms than validator spots? e.g. total ETH staked is the "public resource" (land) that's being shared and extracted utility from. cap it at 35 million. but not really, number of validators is the real problem